such that the hierarchical team is feasible for the principal and he prefers it to a simultaneous team and competing agents.
The natural question which arises in the framework of hierarchical team is whether the team leader prefers a position of a Stackelberg leader or Stackelberg follower in the investment stage of the game. The analysis of the latter situation is untractable in the framework of this model. It is possible to show, however, that the inefficiency resulting from the suboptimal allocation of resources between the leader and the follower is present also in this case.
contributed by one agent makes the other agent more prone to shirk. This result is in line with empirical evidence by Gould and Winter (2005). Investigating the perfor-mance of professional baseball players the authors conclude, that the perforperfor-mance of players with substitutable tasks is negatively correlated and the performance players with complementary tasks is positively correlated.
The last conclusion is related to the tradeoff between hierarchical team and team with-out hierarchy. The hierarchical structure allows the principal to induce the sequential contribution of effort in the environment where he is unable to do it himself. The hier-archical team, however, always leads to the suboptimal allocation of resources, being therefore less efficient, than the simultaneous team. Still, for the range of parame-ters where the agents’ investments are strategic complements, the sequential nature of production process has a positive effect on the incentives. If this positive effect counterbalances the loss of efficiency, the principal can use hierarchical structure to improve performance of the team.
The paper has a number of interesting implications for the organization of production process. It suggests that we should observe principals switching to financing compet-ing teams, rather than a consortium of teams, as the prize in stake increases. The casual examples, such as the one cited in introduction, indicate that this indeed might be the case. Sometimes the necessary degree of competition is provided by the mar-ket, as was the case with Human Genome Project Consortium, the success of which was fuelled by the competition with the private firm Celera.18 But it is easy to see that (at least in the framework of this model) a profit - maximizing principal would prefer to keep competing teams “in house”, rather than rely on the competition with the external team (in which case he is left empty-handed whenever the other team wins).
The paper also implies, that in an environment where a team is organized as a hi-erarchy and a team leader has difficulties verifying the effort of his subordinates, we should observe the team leader executing significantly larger effort, than his team peers. This result is similar to Hermalin (1998) who concludes that a team leader should work harder than his subordinates. His result relies, however, on the fact that the leader has an information which is not available to the other team members and signals this information by “leading by example”. I provide yet another mechanism which explains why team leaders tend to work harder than their subordinates. This mechanism is important if a team leader can decide on the allocation of reward
be-18In 2003 HGP Consortium and Celera simultaneously and independently published the sequence of human genome, seenews.bbc.co.uk/1/hi/sci/tech/2940601.stm
tween himself and his subordinate (which is often the case in R&D). Notice, that in this case unlike in Hermalin (1998), the tendency of the team leader to exert higher effort than his subordinate reduces the efficiency of production.
The obvious question that arises in connection to the results presented in the paper, is what should be the optimal structure of a production process with more than two agents? There is a rich set of possibilities from N competing agents to the multiple levels of hierarchy. The investigation of this interesting problem is a direction for a future research.
Chapter 3
On Compatibility in Two-Sided Markets
3.1 Introduction
The problem of compatibility choice in the framework of markets with network exter-nalities has received much attention in the literature. This is not surprising, since the compatibility of products in this environment affects the size of relevant network and hence the incentives of agents to buy a particular product. Any decision of a firm op-erating in such market, from R&D to the introduction of upgrades, crucially depends on the fact whether its product is compatible with those of a rival or/and with the previous generations of the same product. It is surprising, however, that investigating compatibility choice, the literature did not pay much attention to the fact that many of the markets exhibiting network externalities are two-sided markets.1
Indeed, examples of two-sided markets are numerous. First of all, they include many industries of the classical economy: newspapers and TV-channels, commercial fairs, dating agencies and night clubs, shopping malls, etc. The most prominent examples, however, are related to the New Economy in general and to software platforms in par-ticular. Operating systems, video-game consoles, payment cards, smart phones and PDA’s all share features of two-sided (or, more generally, multi-sided) markets. In a recent book, Evans, Hagiu and Schmalensee (2006) describe multi-sided software plat-forms asinvisible engines that “are in the process of transforming industries ranging from automobiles to home entertainment” (p. vii).
1Rochet and Tirole (forthcoming) define two-sided markets as markets, where one or several platforms enable interaction between two distinct group of agents and the volume of transaction is affected by a price structure.
In this paper we investigate the choice of compatibility between two generations of platforms (old and new) in the framework of two-sided markets. We provide classi-fication of the compatibility regimes that one can observe on two-sided markets and develop a theory that explains how the choice of a particular regime depends on the characteristics of the market (the size of the installed base and the market growth rate) and technological features of the new platform. We show that the driving force which determines the choice of a compatibility regime is the tradeoff between incen-tives of the new agents on one side of the market and the incenincen-tives of the installed base on the other side of the market.
This paper is motivated by two observations. First, compatibility of technologies on two-sided markets has several regimes. Obviously, platforms may be incompatible with each other. GameCube, a video game console of Nintendo, is incompatible with its predecessor, N 64. Further, platforms may be backward compatible for one side of the market. Sony PlayStation 3, for example, is backward compatible with Sony PlayStation 2, its predecessor: a user of the former can play any games designed for the latter. Finally, platforms may be fully compatible with each other, as is the case with Palm OS. Not only a user of Palm OS can run on it any program designed for the older version of this operation system, but also any program designed for the new version of operation system can be run on the older version.
The second observation is that the choice of compatibility not only differs across industries (as illustrated by examples above) but (for the same firm) across time periods. As an example, consider Nintendo, which, after producing generations of incompatible game consoles, made its new game console, Wii, backward compatible with its predecessor, GameCube.
To explain these observations and to provide a theory of compatibility choice in two-sided markets we consider a framework with two platforms owned and operated by a single firm (referred to as the monopolist). The platforms enable interaction between two groups of agents, labelled as users and sellers. One of the platforms represents an old generation of technology and the other platform represents a new generation of technology. The new platform is superior to the old one in the extent of network benefits (which we also callper-interaction benefits) that it confers to users and sellers.
In addition it has some intrinsic benefits, that are independent on the size of network, and reflect fashion or alternative uses of the platform. The size of network benefits and stand-alone benefits determine the extent of technological progress.
The old platform has an installed base: some users and sellers are already subscribed to this platform and can use it to interact with each other. In addition, there is a number of new users and sellers entering the market (their measure represents market
growth rate). These new agents cannot subscribe to the old platform. To interact with agents on the other side of the market they have to, therefore, subscribe to the new platform. This indeed reflects the situation on many markets of interest, where the old generation of the platform (for example, an outdated operation system or an old generation of a game console) is no longer available (unless in a secondary market).
The users and sellers are assumed to be heterogeneous with respect to net costs which they incur when adopting a new platform.
The price-discriminating monopolist earns profit by selling the new platform to the installed base and to the new agents and charging them a subscription fee. In ad-dition the monopolist is free to choose among four compatibility regimes: making the new platform incompatible with the old one, fully compatible, or only backward compatible for agents on one side of the market. In the absence of any form of com-patibility, only agents subscribed to the same generation of platform can interact.
By imposing compatibility, the monopolist enables an interaction between users and sellers subscribed to different generations of platform.
Finally, deciding on compatibility, the monopolist can also determine the quality of interaction between agents, subscribed to the new platform and the agents on the other side of the market, subscribed to the old platform. The minimal quality which the monopolist can choose is zero, which corresponds to the situation where the new platform and the old platform are incompatible. It is assumed that the quality of interaction between agents subscribed to different platforms can never exceed the quality of interaction between agents subscribed to the old platform. In other words, the people who play a game designed for PlayStation 2 on PlayStation 3 can only enjoy the graphic and sound to the extent they would enjoy it using PlayStation 2.
Any quality of interaction between zero and the maximal value corresponds topartial compatibility, because it only confers a part of maximal network benefits to agents on both sides od the market.
Our first crucial result is that the monopolist will never choose partial compatibility.
He either will make technologies incompatible for one side on the market or will make them compatible to the extent that agents can enjoy the maximal network benefits.
This result is new to the literature on network externalities, which up to nowassumed that the compatibility is a yes/no decision (Katz and Shapiro 1985, Katz and Shapiro 1986, Farrell and Saloner 1986, Katz and Shapiro 1992, Doganoglu and Wright 2006),2 although there always was an unease about this assumption (see, for example, Choi
2One exception from this rule is Farrell and Saloner (1992), who assume that compatibility is provided through the use of converter, which can be imperfect. However, the quality of converter in their model is exogenously determined and is not chosen by firms, who provide converter.
1994). Our result provides a justification for this assumption and allows to concentrate our analysis on four extreme compatibility regimes (incompatible platforms, fully compatible platforms and two types of backward compatibility).
Analyzing the choice of the compatibility regime, we showed that at the heart of monopolist’s decision to make technologies compatible is the tradeoff between demand of new agents on one side of the market and demand of the old agents on the other side of the market. In particular, if the monopolist introduces backward compatibility for, say, users, he encourages new users to buy the new platform but discourages the old sellers to do so (direct effect). For illustration consider a case, where monopolist makes platforms backward compatible for users. This improves incentives of new users to buy the new platform. Indeed, now, using this platform, they can access the installed base of sellers. On the other hand, the sellers, who belong to the installed base have less incentives to buy the new platform. Indeed, now, using their old platform they can interact with all users, subscribed to the new platform.
The decrease in the demand of old sellers triggers the decrease in demand of old users and of the new users (feedback effect). The negative feedback effect becomes more important if the technological progress is revolutionary, while the direct positive effect less so. Indeed, if the new platform is very advanced, then the new users have large incentives to buy it even if it does not allow them to access the installed base of sellers.
The compatibility therefore will bring only moderate improvement in their demand.
The tradeoff betweendirect and feedback effects determines which type of compatibil-ity will be chosen on the market. To provide trackable analysis of compatibilcompatibil-ity choice, in the second part of the paper we concentrate on several market structures which are characterized by extreme values of one or several parameters and are observed in reality. These are mature market (the market growth rate is small), emerging market (the installed base is small) and the asymmetric market (the installed base exists only on one side of the market).
We characterize the optimal choice of compatibility for these chosen market struc-tures. As follows from our analysis, the monopolist is more likely to make platforms compatible if the technological progress is moderate. Further, the compatibility for, say, users is likely to be imposed if the installed base of sellers is relatively small, the installed base of users is relatively small and the growth rate of their installed base is moderate.
Although our model is static we are able to provide some intuition about dynamics of compatibility choice as the market develops from emerging to mature or as the monop-olist, who treated his market as one-sided business, embraces a two-sided model. We illustrate our predictions with examples from video game console market and market
for personal digital assistants.
The set-up of our model shares common features with the literature on two-sided markets (Rochet and Tirole 2002, Caillaud and Jullien 2003, Armstrong forthcoming, Armstrong and Wright 2004, Rochet and Tirole forthcoming). Our results, however, are novel to this literature, which up to now typically ignored compatibility. The exception is Doganoglu and Wright (2006), who investigate the incentives of competing firms to make their platforms compatible given that consumers of their products may (or may not) multihome, i.e. subscribe to both platforms. The authors mainly investigate markets with simple network externalities (i.e. there is only one group of agents). They, however, also discuss implications of their model for two-sided markets. The focus of this model is very different from ours. First, the incentives to make platforms compatible stem from competition. Second, the ability of consumers to multihome in their model is the driving force of the result, while in our model this is the tradeoff between incentives of old and new agents. Finally, the authors do not distinguish between different compatibility regimes and view compatibility as a yes/no decision (full compatibility/incompatibility).
The literature on compatibility in the presence of simple network externalities may be divided into two groups. The first group of papers investigate compatibility of technologies on perfectly competitive or oligopolistic market. The incentives of firms to make technologies compatible stem mostly from competition. Katz and Shapiro (1986) show that in a dynamic framework the competing firms have incentives to achieve compatibility of the products in order to soften the price competition on the early stage of the industry development. Kristiansen (1998) shows that compatibility may also be used to reduce the R&D competition at the stage of product introduction.
Katz and Shapiro (1992) study a dynamic model, where consumers entering at each date choose between buying a incumbent technology or waiting until the entrant introduces a more advanced technology. The authors show that, depending on the size of the installed base, market growth rate, and consumers’ beliefs, either entrant or incumbent (but seldom both of them) would prefer to make both technologies compatible.
Unlike this strand of literature, we study the situation where both old and new plat-form (technology) are owned by a monopolist. We do this for two reasons. First, the structure on many industries involving multi-sided markets indeed is monopolistic (or close to monopolistic), for instance, PC operating systems with Microsoft, internet auctions with eBay, etc. Second, we want to analyze the incentives for achieving com-patibility other than those which are related to competition. We show in the paper that incentives of the monopolist to make platforms compatible are determined by
the extent to which he looses the demand on the behalf of the existing agents from one side of the market, which free-ride on the compatibility of platforms for agents on the other side of the market.
The second group of papers in the literature on network externalities is a literature on planned obsolescence. The paper which shares a number of similarities with our model in this literature is Choi (1994). This paper considers a decision of the monopolist in a two-period model. The monopolist sells a technology in the first period, forming an installed base, and a new generation of this technology in the second period. He has a choice between making the technologies compatible or incompatible with each other. Choi (1994) shows that the decision to introduce an incompatible technology crucially depends on the fact whether the monopolist intends to sell this technology to both installed base and new agents or only to new agents. In the former case the monopolist will make technologies incompatible, while in the latter case he will make them compatible. The first strategy (incompatible technology, sell to both groups of agents) is shown to be optimal if the new technology has sufficiently high stand-alone benefits and the first group of agents (installed base) is sufficiently large, compared to the number of new agents.
The intuition, underlying these results, that the tradeoff between the demand of old and new agents, is determinant for the compatibility decision, is similar to ours.
Important difference, however, between this paper and Choi (1994) is that in our framework it is demand of the new agents on one side of the market and the demand of the old agents on the other side of the market which matters for compatibility choice. Further, in the framework of two-sided markets we are able to characterize the richer set of compatibility regimes than Choi (1994). Finally, we also investigate how the choice of compatibility regime depends on the extent of network benefits that the new technology confers to the agents on both sides of the market. Turns out that higher network benefits intensify the negative feedback effect while making the positive direct effect less important. This analysis allows us to predict how the choice of compatibility changes with the technological progress.
The remainder of the paper is organized as follows. In Section 3.2 we describe the setup of the model and provide a classification of compatibility regimes. Section 3.3 analyzes compatibility of platforms under a general demand specification. In Sec-tion 3.4 we introduce assumpSec-tion of linear demand funcSec-tion and investigate three market structures: mature market, emerging market and asymmetric market. In Sec-tion 3.5 we illustrate our predicSec-tions about compatibility choice with two examples.
Section 3.6 concludes. Appendix 3.A contains proofs of all lemmas and propositions.
Figures and tables are given in Appendix 3.B.